DIABETICMedicine
DOI: 10.1111/dme.12895
Research: Epidemiology
Ethnic differences in cross-sectional associations between
impaired glucose regulation, identified by oral glucose
tolerance test or HbA1c values, and cardiovascular disease
in a cohort of European and South Asian origin
S. V. Eastwood1, T. Tillin1, J. Mayet2, D. K. Shibata3, A. Wright4, J. Heasman4, N. Beauchamp3,
N. G. Forouhi5, A. D. Hughes1 and N. Chaturvedi1
1
UCL Institute of Cardiovascular Science, University College London, London, 2National Heart and Lung Institute, Imperial College London, London, 3Department
of Radiology, University of Washington Medical Centre, Seattle, WA, USA, 4Department of Radiology, Imperial College NHS Healthcare Trust, London, UK and
5
MRC Epidemiology Unit, University of Cambridge, Cambridge, UK
Accepted 24 August 2015
Abstract
Aims We contrasted impaired glucose regulation (prediabetes) prevalence, defined according to oral glucose tolerance
test or HbA1c values, and studied cross-sectional associations between prediabetes and subclinical/clinical cardiovascular
disease (CVD) in a cohort of European and South Asian origin.
Methods For 682 European and 520 South Asian men and women, aged 58–85 years, glycaemic status was determined
by oral glucose tolerance test or HbA1c thresholds. Questionnaires, record review, coronary artery calcification scores and
cerebral magnetic resonance imaging established clinical plus subclinical coronary heart and cerebrovascular disease.
Prediabetes was more prevalent in South Asian participants when defined by HbA1c rather than by oral glucose
tolerance test criteria. Accounting for age, sex, smoking, systolic blood pressure, triglycerides and waist–hip ratio,
prediabetes was associated with coronary heart disease and cerebrovascular disease in European participants, most
obviously when defined by HbA1c rather than by oral glucose tolerance test [odds ratios for HbA1c-defined prediabetes
1.60 (95% CI 1.07, 2.39) for coronary heart disease and 1.57 (95% CI 1.00, 2.51) for cerebrovascular disease]. By
contrast, non-significant associations were present between oral glucose tolerance test-defined prediabetes only and
coronary heart disease [odds ratio 1.41 (95% CI 0.84, 2.36)] and HbA1c-defined prediabetes only and cerebrovascular
disease [odds ratio 1.39 (95% CI 0.69, 2.78)] in South Asian participants. Prediabetes defined by HbA1c or oral glucose
tolerance test criteria was associated with cardiovascular disease (defined as coronary heart and/or cerebrovascular
disease) in Europeans [odds ratio 1.95 (95% CI 1.31, 2.91) for HbA1c prediabetes criteria] but not in South Asian
participants [odds ratio 1.00 (95% CI 0.62, 2.66); ethnicity interaction P = 0.04].
Results
Conclusions Prediabetes appeared to be less associated with cardiovascular disease in the South Asian than in the
European group. These findings have implications for screening, and early cardiovascular prevention strategies in South
Asian populations.
Diabet. Med. 33, 340–347 (2016)
Introduction
In parallel to the global diabetes epidemic, population surveys
indicate a burgeoning prevalence of impaired glucose regula-
Correspondence to: Sophie V. Eastwood. E-mail: [email protected].
This is an open access article under the terms of the Creative Commons
Attribution License, which permits use, distribution and reproduction in any
medium, provided the original work is properly cited.
340
tion, often termed prediabetes hyperglycaemia (12–29%,
depending on definition) [1]. Although it is clear that the term
‘prediabetes’ is misleading, as not all those with prediabetes
will develop diabetes, it is acknowledged that this may
represent a high cardiovascular disease risk state, with
implications for intervention [2]. It is in that latter sense that
we use this term. Conventionally, prediabetic states have been
defined by either post-glucose challenge [impaired glucose
tolerance (IGT)] or fasting glycaemia [impaired fasting
ª 2015 The Authors.
Diabetic Medicine published by John Wiley & Sons Ltd on behalf of Diabetes UK.
Research article
What’s new?
• For participants of European origin, HbA1c values
defined as impaired glucose regulation (prediabetes)
were cross-sectionally associated with coronary heart
disease, cerebrovascular disease and the composite
outcome of cardiovascular disease.
• By contrast, in South Asian participants prediabetes
defined according to HbA1c concentration was nonsignificantly associated with cerebrovascular disease
only, and associations between HbA1c concentrationdefined prediabetes and overall cardiovascular disease
were absent and, thus significantly weaker, than those
seen in the European participants.
• This suggests that current prediabetes HbA1c thresholds
may be inappropriate in South Asian groups as prediabetes defined in this way did not appear to confer
excess cardiovascular risk in this ethnic group.
glycaemia (IFG)] [3], but recently, guidelines for HbA1c-based
definitions of prediabetes have been published [2,4].
Relatively more prediabetes is identified by HbA1c than by
IFG/IGT criteria in people of South Asian origin, whereas
prevalence is similar by either criterion in populations of
European origin [5]. It is unclear how the greater prevalence
of HbA1c-identified prediabetes in this former group translates to cardiovascular risk, with some authors suggesting the
HbA1c definition of prediabetes may be less discriminative
[6]. As far as we are aware, this has never before been studied
in South Asian populations; the majority of studies that have
compared associations between prediabetes and cardiovascular disease by diagnostic criteria are in populations of
European origin [7–10]. Moreover, our previous work
suggests that associations between diabetes and cardiovascular disease (CVD) vary with ethnicity, with stronger
associations in South Asian than in European populations
[9,11]. Whether similar ethnic differences exist for associations between prediabetes and CVD risk is unclear.
The aims of the present study were to establish whether
associations between prediabetes and subclinical and clinical
cardiovascular disease risk differ between UK European and
South Asian ethnic groups and whether any ethnic differences in associations were related to the definition of
prediabetes used.
Subjects and methods
We used cross-sectional data from the Southall and Brent
Revisited (SABRE) study, a multi-ethnic population-based
cohort of individuals living in north-west London [12]. The
South Asian participants were first-generation migrants;
82% were born in the Indian subcontinent and 14% in East
Africa. Participants aged 40–70 years (n = 4056) were
ª 2015 The Authors.
Diabetic Medicine published by John Wiley & Sons Ltd on behalf of Diabetes UK.
DIABETICMedicine
randomly selected from age- and gender-stratified general
practitioner lists and workplaces at baseline (1988–1991),
and followed up (2008–2011) when aged 58-85 years
(n = 2671). The present analysis concerns the 682 European
and 520 South Asian participants who attended the followup clinic. All participants gave written informed consent.
Study approval was obtained from St Mary’s Hospital
Research Ethics Committee (07/H0712/109).
Smoking status, alcohol consumption and medication
receipt were ascertained from a health and lifestyle questionnaire [12]. Physical activity comprised the total weekly
energy expenditure (MJ), as previously described [13]. Blood
pressure was obtained three times after a 15-min rest with an
Omron CEP 7050 (Omron, Tokyo, Japan); the mean of the
final two readings was used in the analysis. Glucose levels,
lipid profile, HbA1c and C-reactive protein levels were
measured on fasting blood samples, and anthropometry
performed [13,14]. In addition, an abdominal computed
tomography (CT) slice was taken to measure visceral adipose
tissue area [13]. Participants without known diabetes
underwent an oral glucose tolerance test (OGTT).
Cardiac CT scanning was performed in all participants
from the ascending aorta above the level of the coronary
arteries to the inferior border of the heart. Coronary artery
calcification was quantified using proprietary software on a
Phillips Extended Brilliance computer workstation (Phillips
Healthcare, Eindhoven, The Netherlands), and calcification
was defined as an area >1 mm2 of density >130 Hounsfield
units. The coronary artery calcification score was calculated
as the sum of all lesion scores [Agatston units (AU)]. Scans
were read by a single experienced observer blinded to
participant ethnicity. Interobserver reproducibility, comparing scores from a senior investigator (A.W.) and
intra-observer reproducibility were assessed initially and at
intervals during follow-up, using the same 20 CT scans. The
intra-class correlation coefficient for intra- and inter-observer
measurements was 0.94. Subclinical coronary heart disease
(CHD) was classified as a coronary artery calcification score
>400 AU [15].
Cerebral magnetic resonance imaging (MRI) scans provided data on brain infarcts ≥3 mm, a subclinical measure of
cerebrovascular disease. We used an MRI scanning and
scoring protocol based on that of the Cardiovascular Health
Study [16]. Whole-brain scans included sagittal T1-weighted
images and axial T1-weighted, proton density and
T2-weighted images of 5-mm thickness, with no gaps.
Thin-section 3-mm axial fluid attenuated inversion recovery
(FLAIR) and coronal 1.5-mm three-dimensional T1-weighted
gradient echo images were also obtained. Scans were
performed using a General Electric 1.5T or 3T scanners.
Only infarcts of ≥3 mm were assessed, as smaller lesions are
less reproducible [14]. Inter- and intra-observer reproducibility were evaluated for the presence of brain infarcts
≥3 mm on 44 scans; inter-observer j was 0.68 and intraobserver j was 0.79.
341
DIABETICMedicine
Impaired glucose regulation, cardiovascular disease and ethnicity S. V. Eastwood et al.
We used three classification systems to define glycaemic
status for participants without existing diabetes. Firstly,
WHO 1999 criteria were used to define prediabetes [either
impaired fasting glycaemia (fasting glucose ≥6.1 mmol/l and
<7.0 mmol/l) or impaired glucose tolerance (2-h OGTT
plasma glucose ≥7.8 mmol/l and <11.1 mmol/l)] and new
diabetes (fasting glucose ≥7.0 mmol/l or 2-h OGTT plasma
glucose ≥11.1 mmol/l] [3]. Secondly, glycaemic categories
according to the International Expert Committee (IEC) 2009
criteria [2] were based on the following HbA1c thresholds:
prediabetes, HbA1c ≥42 mmol/mol (6.0%) but < 48 mmol/
mol (6.5%); new diabetes, HbA1c ≥48 mmol/mol (6.5%).
We also studied glycaemia according to the American
Diabetes Association (ADA) 2014 recommendations [4],
which advocate HbA1c thresholds of ≥39 mmol/mol (5.7%)
but <48 mmol/mol (6.5%) for prediabetes and ≥48 mmol/
mol (6.5%) for new diabetes. Analyses were conducted using
the IEC HbA1c prediabetes threshold unless otherwise stated.
Pre-existing diabetes was identified from primary care record
review or participant questionnaire (recall of physiciandiagnosed diabetes plus diagnosis year or diabetes medication).
Coronary heart disease was defined firstly from primary
care record review adjudicated by two clinicians, as per
Anglo-Scandinavian Cardiac Outcomes Trial (ASCOT) criteria [17]. Additionally, International Classification of
Disease (ICD)-9 codes 410-415 and ICD-10 codes I200I259 from Hospital Episode Statistics and codes K401-K469,
K491-K504, K751-K759 and U541 from the Office of
Populations and Surveys classification of interventions and
procedures identified CHD. For stroke, primary care data
were reviewed in a similar manner to that used for CHD,
with diagnoses made again according to ASCOT criteria
[17]. Stroke was also ascertained from participant report of
physician-diagnosed stroke with duration of symptoms
≥ 24 h and from Hospital Episode Statistics (diagnostic
ICD-9 codes 430 to 439 or ICD-10 codes I600 to I698).
We examined descriptive statistics for demographics,
cardiometabolic risk factors, medication use and CVD
outcomes by ethnicity and glycaemic status. Logistic and
linear regression techniques determined age- and sex-adjusted group differences. Glycaemic status by OGTT or
HbA1c criteria was contrasted graphically in each ethnic
group, using OGTT then HbA1c categories in turn as the
referent groups. To enhance power and address potential
ethnic presentation biases, we combined established and
subclinical disease outcomes for CHD and cerebrovascular
disease (subclinical disease defined as coronary artery calcification >400 AU and brain infarct ≥3 mm, respectively). In
addition, we created a composite CVD prevalence outcome
(comprising CHD or cerebrovascular disease). Each outcome
was examined by glycaemic status and ethnicity.
Following this, associations between glycaemic status and
CHD, cerebrovascular disease or CVD were evaluated using
age- and sex-adjusted logistic regression models within each
342
ethnic group. We then further adjusted these models for
potential confounders (smoking, systolic blood pressure,
triglycerides and waist–hip ratio), selected on the basis of
significant associations with any CVD indicator and either
prediabetes measure. We explored replacing waist–hip ratio
with the more direct measure of visceral adipose tissue on CT
in multivariate models, but this made little qualitative
difference to our findings, so we present data adjusted for
waist–hip ratio as the more familiar measure. To gauge the
relative merit of each prediabetes measure for the prediction
of CVD in each ethnic group, we compared C-statistics for
age- and sex-adjusted models of CVD with models with the
addition of either prediabetes measure.
Analyses were repeated using ADA HbA1c criteria for
prediabetes. We further adjusted multivariable models for
medication use (as a sensitivity analysis because of data
sparsity). Analyses were conducted in STATA 13 (College
Station, TX, USA), with P values <0.05 taken to indicate
statistical significance.
Results
Overt diabetes was more prevalent, and waist–hip ratio and
systolic blood pressure higher, while HDL cholesterol was
lower in the South Asian group than in the European group
(Table 1). Irrespective of ethnic group, increasing glycaemia
was associated with worsening of cardiometabolic risk
factors (Table S1).
In the South Asian but not the European participants, there
was a 1.5-fold increase in the prevalence of prediabetes when
HbA1c rather than OGTT criteria were used (Table 1). When
ADA HbA1c thresholds were used rather than OGTT
criteria, the prevalence of prediabetes doubled for both
European and South Asian participants (Table S2). South
Asian participants had more clinical CHD and overall CVD
(CHD + cerebrovascular disease) than European participants, although there were no differences between the ethnic
groups in the prevalence of subclinical CHD or subclinical
and clinical cerebrovascular disease (Table 1).
Differences in classification were examined by comparing
glycaemic status according to either diagnostic criterion,
using OGTT categories as the index classification, and
excluding those with existing diabetes (Fig. 1a). With the
exception of normoglycaemia in Europeans, agreement was
poor (40–49%) for all categories in all groups, with a
tendency for South Asian participants to be placed in a
more adverse glycaemic category by HbA1c classification.
When HbA1c categories were used as the reference
(Fig. 1b), there was good agreement (71–75%) for normoglycaemia in both ethnic groups, but poor agreement for
prediabetes and diabetes, especially for South Asian participants.
In the European group, CHD, cerebrovascular disease and
CVD were more prevalent in prediabetes (by either criterion)
than participants with normoglycaemia (Table 2). A similar,
ª 2015 The Authors.
Diabetic Medicine published by John Wiley & Sons Ltd on behalf of Diabetes UK.
Research article
DIABETICMedicine
Table 1 Characteristics of participants in the Southall and Brent Revisited study, by ethnicity
Number of participants
Median (IQR) age, years
Female sex, n (%)
Ever smoked
Median (IQR) alcohol consumption, units/ week
Median (IQR) physical activity, MJ/ week
Mean SD waist–hip ratio
Mean sd BMI, kg/m2
Mean SD systolic blood pressure, mmHg
Mean sd diastolic blood pressure, mmHg
Mean SD HDL cholesterol, mmol/l
Median (IQR) triglycerides, mmol/l
Median (IQR) C-reactive protein, mmol/l
Anti-hypertensive medication, n (%)
Lipid-lowering medication, n (%)
Glycaemic status: OGTT, n (%)
Normoglycaemia
Prediabetes
Diabetes*
Glycaemic status: HbA1c, n (%)
Normoglycaemia
Prediabetes
Diabetes*
CHD, n (%)
Clinical CHD
Coronary artery calcification >400 AU
Clinical CHD or coronary artery calcification‡ >400 AU
Cerebrovascular disease, n (%)
Stroke
Brain infarct ≥3 mm
Stroke or brain infarct ≥3 mm
CHD or cerebrovascular disease, n (%)
European ethnicity
South Asian ethnicity
P†
682
70 (65–75)
153 (22)
426 (63)
5 (1–14)
9.4 (6.8–12.1)
0.97 0.07
28 5
138 17
77 10
1.4 0.4
1.1 (0.9–1.6)
1.7 (0.8–3.6)
371 (54)
330 (48)
520
68 (64–73)
78 (15)
113 (22)
4 (1–8)
8.8 (5.8–11.7)
1.00 0.07
26 4
142 18
76 10
1.3 0.3
1.2 (0.9–1.7)
1.4 (0.7–3.0)
395 (76)
354 (68)
–
0.006
0.001
<0.001
<0.001
0.001
<0.001
<0.001
<0.001
0.009
0.001
0.09
0.01
<0.001
<0.001
377 (55)
172 (25)
133 (20)
201 (39)
99 (19)
220 (42)
<0.001
0.01
<0.001
374 (55)
177 (26)
131 (19)
136 (26)
146 (28)
238 (46)
<0.001
0.37
<0.001
142 (21)
119 (23)
261 (38)
189 (36)
72 (22)
258 (50)
<0.001
0.79
<0.001
32
126
141
335
25
94
105
296
(5)
(21)
(23)
(49)
(5)
(19)
(21)
(57)
0.68
0.92
0.94
0.001
AU, Agatston units; CHD, coronary heart disease; IFG, impaired fasting glucose; IGT, impaired glucose tolerance; IQR, interquartile range;
OGTT, oral glucose tolerance test.
*Includes pre-existing and newly diagnosed diabetes. Prediabetes by OGTT criteria comprised IFG and/or IGT.
†
Age and sex-adjusted P value for ethnic difference, when compared with European participants.
‡
Measured in participants without known CHD.
but less pronounced, pattern was seen in South Asian
participants. This was even less marked when we examined
disease prevalence by ADA HbA1c thresholds, in both the
European and South Asian groups (Table S3). When
subclinical and clinical CVD measures were examined
separately, similar patterns were seen (Table S4). Notably,
the prevalence of clinical CHD in the South Asian participants with normoglycaemia was similar to that observed in
the European participants with prediabetes.
In age- and sex-adjusted models of CHD or cerebrovascular disease, prediabetes by either criterion appeared to be
related to disease risk in the European group (Table 3).
Accounting for smoking, systolic blood pressure, triglycerides and waist–hip ratio attenuated these associations, but
risks remained elevated for people with prediabetes and
statistical significance was retained for HbA1c-defined prediabetes. Prediabetes by either criterion was associated with
overall CVD, with associations being of higher magnitude
when defined by HbA1c rather than OGTT. By contrast, in
South Asian participants, non-significant associations were
ª 2015 The Authors.
Diabetic Medicine published by John Wiley & Sons Ltd on behalf of Diabetes UK.
present between OGTT-defined prediabetes only and CHD
[odds ratio 1.39 (95% CI 0.84, 2.29)] and HbA1c-defined
prediabetes only and cerebrovascular disease [odds ratio 1.41
(95% CI 0.72, 2.76)]). Prediabetes by either criterion
appeared unrelated to CVD in this group. Associations
between prediabetes by HbA1c and CVD were greater in the
European than in the South Asian group (ethnicity interaction P = 0.04, respectively). Analyses using ADA HbA1c
thresholds showed similar patterns to the main results
(Table S5).
When either measure of prediabetes was added to age- and
sex-adjusted models of CVD, there were no improvements in
the C-statistic for South Asian participants; however, addition of either OGTT- or HbA1c-defined prediabetes to
models marginally improved the C-statistic in Europeans,
more so for the latter [from 0.695 (95% CI 0.632, 0.758) to
0.705 (95% CI 0.642, 0.767)] with addition of OGTTdefined impaired glucose regulation (P = 0.20) or 0.708
(95% CI 0.646, 0.769] with addition of HbA1c-defined
impaired glucose regulation (P = 0.008).
343
Impaired glucose regulation, cardiovascular disease and ethnicity S. V. Eastwood et al.
DIABETICMedicine
(a)
HbA1c defintions
normoglycaemia
100
1%
8%
pre-diabetes
7%
90
% categorised by HbA1c glycaemic categories
diabetes
24%
26%
80
42%
46%
40%
70
44%
60
40%
50
32%
40
73%
44%
30
48%
53%
20
36%
26%
10
10%
0
Europeans
South Asians
Normoglycaemia
Europeans South Asians
Europeans South Asians
Diabetes
Pre-diabetes: IFG/IGT
Glycaemic categories by fasting/post-load glucose
(b)
Fasting/post-load glucose definitions
pre-diabetes (IFG/IGT)
% categorised by fasting/post-load glycaemic categories
normoglycaemia
100
2%
3%
25%
26%
diabetes
6%
12%
90
32%
80
38%
27%
45%
70
60
50
40%
40
71%
73%
30
41%
56%
61%
20
28%
10
14%
0
Europeans
South Asians
Normoglycaemia
Europeans South Asians
Pre-diabetes
Europeans
South Asians
Diabetes
Glycaemic categories by HbA1c
FIGURE 1 (a) Glycaemia categorized by oral glucose tolerance test
(OGTT) vs. HbA1c, by ethnicity. Diabetes comprises newly diagnosed
cases only. (b) Glycaemia categorized by HbA1c vs. OGTT, by
ethnicity. Diabetes comprised newly diagnosed cases only.
Finally, further adjustment of the multivariable models for
receipt of antihypertensive or lipid-lowering medication
(Table S6) did not alter the main results.
Discussion
There was a marked ethnic variation in prediabetes by
different diagnostic criteria [2–4], with a higher prevalence in
South Asian but not European participants, when prediabetes was defined by HbA1c rather than OGTT criteria.
Furthermore, prediabetes appeared to relate differently to
cardiovascular disease according to diagnostic criteria and
ethnicity; HbA1c-defined prediabetes was associated with
CHD and cerebrovascular disease in the European group, but
344
with only cerebrovascular disease (non-significantly) in the
South Asian group. Furthermore, there was some evidence of
a weak association between OGTT-defined prediabetes and
CHD in the European and South Asian groups. Prediabetes
by either criterion was associated with total CVD (CHD +
cerebrovascular disease) in the European but not the South
Asian group.
When HbA1c, as opposed to OGTT, criteria were used to
define prediabetes, South Asian participants had a higher
prevalence of prediabetes, reflecting previous work [5]. This
may indicate ethnic differences in the propensity to glycate
haemoglobin [18] or in pathways to overt hyperglycaemia
[19].
Reflecting established ethnic differences in CVD, the South
Asian group had greater clinical CHD than did the European
group [11]; however, proportions with subclinical CHD were
similar in European and South Asian participants, which
might be explained by our previous finding that coronary
artery calcification scores are less related to coronary artery
stenosis on angiography in South Asian than in European
people [20]. Also of note, South Asian participants in the
normoglycaemic category (by any criteria) had clinical CHD
rates akin to those of the European participants in the
prediabetes category, implying that even normoglycaemia is
a relatively high risk state in South Asian people, and thus
risk factor management at the prediabetes stage may be too
late.
In the European participants, associations between prediabetes by either criteria and CHD support the argument for
preventative interventions at this stage of glycaemia and are
consistent with previous studies reporting the effects of IFG/
IGT [7, 21, 22] and HbA1c [23]. This was less clear for the
South Asian participants, for whom only IFG/IGT-defined
prediabetes appeared detrimental. We are unaware of any
studies comparing associations between prediabetes and
CHD risk in European and South Asian participants,
although a study of South Asian and European participants
living in Canada found similar associations between HbA1c
and carotid intima media thickness [24]. The underlying
explanations for ethnic differences in the effects of glycaemia
are not clear, and probably involve a combination of genetic,
epigenetic and lifestyle factors.
Associations between HbA1c-defined prediabetes and cerebrovascular disease appeared similar in the two ethnic
groups, although OGTT-defined prediabetes was associated
(albeit weakly) with cerebrovascular disease in the European
participants only; again, explanations are not obvious. A
meta-analysis of studies examining associations between IFG
or IGT and stroke suggested that IGT was modestly
associated with stroke [25], but a study comparing associations between IFG or HbA1c-defined prediabetes in white
and black populations in the USA found that HbA1c-defined
prediabetes, but not IFG, was adversely related to stroke risk
[26]. This latter study reflects our findings to some extent,
but we used more stringent HbA1c thresholds (IEC, as
ª 2015 The Authors.
Diabetic Medicine published by John Wiley & Sons Ltd on behalf of Diabetes UK.
Research article
DIABETICMedicine
Table 2 Cardiovascular disease prevalence by ethnicity and glycaemic status in the Southall and Brent Revisited study
Atherosclerosis measure
CHD: clinical or coronary
artery calcification
score >400 AU
Cerebrovascular disease:
stroke or brain infarct ≥3 mm
CHD or cerebrovascular
disease
European group, n (%)
Glycaemia
measure
Normoglycaemia
OGTT
HbA1c
116 (31)
104 (28)
OGTT
HbA1c
OGTT
HbA1c
70 (21)
63 (18)
155 (41)
140 (37)
South Asian group, n (%)
Prediabetes
71 (41)*
78 (44)**
41
47
94
106
(26)*
(31)*
(55)**
(60)***
Diabetes
Normoglycaemia
Prediabetes
Diabetes
74 (56)***
79 (60)***
76 (38)
49 (36)
47 (47)
60 (41)
135 (61)***
149 (63)***
30
31
86
89
29
18
90
60
15
28
51
71
61
59
155
165
(27)
(27)
(65)***
(68)***
(15)
(14)
(45)
(44)
(16)
(20)
(52)
(49)
(31)**
(27)*
(70)***
(69)***
AU, Agatston units; CHD, coronary heart disease; IFG, impaired fasting glucose; IGT, impaired glucose tolerance; OGTT, oral glucose
tolerance test.
Sex- and age-adjusted P value for difference between prediabetes or diabetes and normoglycaemia: ***p < 0.001, **p < 0.01, *p < 0.05.
Diabetes includes pre-existing and newly diagnosed diabetes by the relevant criterion. Prediabetes by OGTT criteria comprised IFG and/or
IGT.
Table 3 Multivariable models of cardiovascular disease by ethnicity and glycaemic status in the Southall and Brent Revisited study
Glycaemia
measure
European group
Model
Normoglycaemia
South Asian group
Prediabetes
CHD (clinical or coronary artery calcification score >400 AU)
OGTT
1
1
1.55 (1.05,2.29)*
2
1
1.46 (0.98,2.19)
1
1
1.70 (1.15,2.51)**
HbA1c
2
1
1.60 (1.07,2.39)*
Cerebrovascular disease (stroke or brain infarct≥3 mm)
OGTT
1
1
1.42 (0.90,2.25)
2
1
1.25 (0.63,1.88)
1
1
1.73 (1.10, 2.72)*
HbA1c
2
1
1.57 (1.00, 2.51)*
CHD or cerebrovascular disease
OGTT
1
1
1.72 (1.17, 2.54)**
2
1
1.58 (1.07, 2.36)*
1
1
2.10 (1.43, 3.08)***
HbA1c
2
1
1.95 (1.31, 2.91)**
Diabetes
Normoglycaemia
Prediabetes
Diabetes
2.86
2.53
3.79
3.42
(1.87,4.38)***
(1.61,3.98)***
(2.45,5.86)***
(2.15,5.42)***
1
1
1
1
1.39
1.41
1.15
1.06
(0.84,2.29)
(0.84,2.36)
(0.70,1.90)
(0.64,1.77)
2.53
2.26
2.88
2.50
(1.69,3.79)***
(1.48,3.45)***
(1.84,4.53)***
(1.56,4.02)***
1.28
1.09
1.49
1.29
(0.77,2.13)
(0.63,1.95)
(0.90, 2.49)
(0.75,2.22)
1
1
1
1
0.91
1.02
1.41
1.39
(0.45,1.86)
(0.49,4.44)
(0.72,2.76)
(0.69,2.78)
2.49
2.58
2.18
2.22
(1.48,
(1.49,
(1.19,
(1.17,
4.18)**
4.44)**
4.00)*
4.22)*
2.60
2.35
3.39
3.15
(1.68,
(1.47,
(2.18,
(1.96,
1
1
1
1
1.19
1.20
1.10
1.00
(0.72,1.98)
(0.71, 2.02)
(0.67,1.79)†
(0.61,1.66)†
2.89
2.64
2.79
2.63
(1.91,
(1.71,
(1.77,
(1.64,
4.39)***
4.08)***
4.39)***
4.22)***
4.00)***
3.73)***
5.28)***
5.05)***
AU, Agatston units; CHD, coronary heart disease; IFG, impaired fasting glucose; IGT, impaired glucose tolerance; OGTT, oral glucose
tolerance test.
Data are odds ratios (95% CI) for the presence of cardiovascular disease.
***P < 0.001, **P < 0.01, *P < 0.05 for difference between prediabetes or diabetes and normoglycaemia.
†P < 0.05 for ethnic difference, when compared with Europeans. Diabetes includes pre-existing and newly diagnosed diabetes by the relevant
criterion. Prediabetes by OGTT criteria comprised IFG and/or IGT. Model 1: adjusted for age + sex; model 2: adjusted for age + sex +
smoking status + systolic blood pressure + triglycerides (log-transformed)) + waist–hip ratio.
opposed to ADA criteria), more representative of UK clinical
practice [27]. As far as we know, there are no studies
comparing the effect of prediabetes on cerebrovascular
disease in European and South Asian cohorts.
Importantly, OGTT-, and in particular, HbA1c-defined
prediabetes appeared to be related to overall CVD risk in
the European but not in the South Asian group. This
suggests that prediabetes may be a less useful clinical
indicator of overall CVD risk in this group, congruent with
both the findings of our model discrimination analyses and
concerns other authors have expressed [6]. Explanations for
this difference for HbA1c-based prediabetes may lie in the
greater proportions identified by this measure in these
ª 2015 The Authors.
Diabetic Medicine published by John Wiley & Sons Ltd on behalf of Diabetes UK.
ethnic groups; that is, there is less of a differential, in terms
of cardiometabolic profile, between individuals with normoglycaemia and those with prediabetes, although obviously this does not apply to OGTT-identified individuals
with prediabetes. Alternatively, the loss of differential in
risk may come in the normoglycaemic group, for South
Asian participants at least, in whom we showed a similar
prevalence of CHD to that in European participants in the
prediabetic state. Our observation that associations
between prediabetes and CVD were weaker in the South
Asian group than in the European group, whilst associations between diabetes itself and CVD were stronger in the
South Asian group than in the European group, may
345
DIABETICMedicine
Impaired glucose regulation, cardiovascular disease and ethnicity S. V. Eastwood et al.
suggest that South Asian people destined to transition to
diabetes move faster through the prediabetes state than do
European people. This speculation has support from a
recent publication, suggesting that the age-related trajectory
in fasting glucose is greater in South Asian people without
diabetes than in European people [28]. Regardless, these
findings, if substantiated by future studies, have crucial
policy implications for prediabetes screening in ethnic
minority groups.
Using the ADA rather than the IEC criteria to define
prediabetes showed similar trends and ethnic differences in
the association with CHD, stroke and CVD, although odds
ratios were often weaker, probably reflecting the lower
HbA1c threshold for ADA-defined prediabetes.
The strengths of the present study include the high
prevalence of CVD outcomes. Additionally, by capturing
subclinical and clinical disease, we avoided presentation bias,
which may operate when comparing different ethnic groups.
There were no ethnic differences in survivorship (P = 0.32)
or follow-up participation (P = 0.62) of the cohort, suggesting survivor bias is unlikely to have greatly affected the
ethnic differences we found. Furthermore, numbers of events
were similar in each group, therefore, differences in power by
ethnicity are unlikely to explain the lack of key associations
in the South Asian participants. High rates of medication
usage in this elderly population may have altered associations, although adjustment for medication use did not greatly
alter results and the inclusion of participants in receipt of
medication is likely to render the findings more generalizable
to the elderly population as a whole. Measurement error may
have been a source of bias, particularly regarding glucose and
HbA1c measurement; there is not insubstantial biological
variation in both measures [4]. The present study only
examined cross-sectional associations; therefore, we cannot
infer causality and further longitudinal research is required to
replicate the findings.
In summary, the choice of diagnostic criterion for prediabetes markedly influences both the number of individuals
identified in the South Asian group and the CVD risk
conferred by prediabetes. When considering overall CVD
risk, prediabetes was associated with risk in the European
but not in the South Asian group; therefore, the use of
prediabetes as a marker of CVD risk may be less clinically
meaningful in this group. Of importance when considering
CHD risk estimation is the observation that South Asian
individuals with normoglycaemia have similar risks of CHD
to European individuals with prediabetes. These novel
findings have implications for CVD risk stratification and
targeting of interventions in different ethnic groups, and
further work is needed to substantiate them.
Funding sources
The study was funded at baseline by the UK Medical
Research Council, Diabetes UK and the British Heart
346
Foundation and at follow-up by the Wellcome Trust and
British Heart Foundation. The authors also wish to acknowledge the NIHR Imperial Biomedical Research Centre for
structural support regarding the data collection at Imperial
Trust NHS Campus and MRC Epidemiology Unit core
support (NGF) MC_UU_12015/5. Funders played no role in
the study design, conduct or analysis, or in the decision to
submit the manuscript for publication. The SABRE study
group is entirely independent from the funding bodies.
Competing interests
None declared.
Acknowledgements
The authors would like to thank all members of the SABRE
group for their contributions to study design, management,
data collection and analyses.
References
1 Selvin E, Parrinello CM, Sacks DB, Coresh J. Trends in prevalence
and control of diabetes in the United States, 1988-1994 and 19992010. Ann Intern Med 2014; 160: 517–525.
2 International Expert Committee report on the role of the A1C assay
in the diagnosis of diabetes. Diabetes Care 2009; 32:1327–1334.
3 World Health Organisation. Definition, Diagnosis and Classification of Diabetes Mellitus and its Complications. Part 1: Diagnosis
and Classification of Diabetes Mellitus. Geneva: WHO, 1999.
4 Diagnosis and classification of diabetes mellitus. Diabetes Care
2014; 37(Suppl. 1): S81–S90.
5 Mostafa SA, Davies MJ, Webb D, Gray LJ, Srinivasan BT, Jarvis J
et al. The potential impact of using glycated haemoglobin as the
preferred diagnostic tool for detecting Type 2 diabetes mellitus.
Diabet Med 2010; 27: 762–769.
6 Likhari T, Gama R. Ethnic differences in glycated haemoglobin
between white subjects and those of South Asian origin with normal
glucose tolerance. J Clin Pathol 2010; 63: 278–280.
7 de Vegt F, Dekker JM, Ruhe HG, Stehouwer CD, Nijpels G, Bouter
LM et al. Hyperglycaemia is associated with all-cause and cardiovascular mortality in the Hoorn population: the Hoorn Study.
Diabetologia 1999; 42: 926–931.
8 Marini MA, Succurro E, Castaldo E, Cufone S, Arturi F, Sciacqua A
et al. Cardiometabolic risk profiles and carotid atherosclerosis in
individuals with prediabetes identified by fasting glucose, postchallenge glucose, and hemoglobin A1c criteria. Diabetes Care 2012;
35: 1144–1149.
9 Park CM, Tillin T, March K, Ghosh AK, Jones S, Wright A et al.
Hyperglycemia has a greater impact on left ventricle function in
South Asians than in Europeans. Diabetes Care 2014; 37: 1124–
1131.
10 Qiao Q, Dekker JM, de Vegt F, Nijpels G, Nissinen A, Stehouwer
CD et al. Two prospective studies found that elevated 2-hr glucose
predicted male mortality independent of fasting glucose and
HbA1c. J Clin Epidemiol 2004; 57: 590–596.
11 Tillin T, Hughes AD, Mayet J, Whincup P, Sattar N, Forouhi NG
et al. The relationship between metabolic risk factors and incident
cardiovascular disease in Europeans, South Asians, and African
Caribbeans: SABRE (Southall and Brent Revisited) – a prospective
population-based study. J Am Coll Cardiol 2013; 61: 1777–1786.
ª 2015 The Authors.
Diabetic Medicine published by John Wiley & Sons Ltd on behalf of Diabetes UK.
Research article
12 Tillin T, Forouhi NG, McKeigue PM, Chaturvedi N. Southall And
Brent REvisited: Cohort profile of SABRE, a UK population-based
comparison of cardiovascular disease and diabetes in people of
European, Indian Asian and African Caribbean origins. Int J
Epidemiol 2012; 41: 33–42.
13 Eastwood SV, Tillin T, Wright A, Mayet J, Godsland I, Forouhi
NG et al. Thigh fat and muscle each contribute to excess
cardiometabolic risk in South Asians, independent of visceral
adipose tissue. Obesity 2014; 22: 2071–2079.
14 Shibata D, Tillin T, Beauchamp N, Heasman J, Hughes AD, Park C
et al. African Caribbeans have greater subclinical cerebrovascular
disease than Europeans: this is associated with both their elevated
resting and ambulatory blood pressure and their hyperglycaemia. J
Hypertension 2013; 31: 2391–2399.
15 Rumberger JA, Brundage BH, Rader DJ, Kondos G. Electron beam
computed tomographic coronary calcium scanning: a review and
guidelines for use in asymptomatic persons. Mayo Clin Proc 1999;
74: 243–252.
16 Bryan RN, Manolio TA, Schertz LD, Jungreis C, Poirier VC, Elster
AD et al. A method for using MR to evaluate the effects of
cardiovascular disease on the brain: the cardiovascular health
study. AJNR Am J Neuroradiol 1994; 15: 1625–1633.
17 Sever PS, Dahlof B, Poulter NR, Wedel H, Beevers G, Caulfield M
et al. Rationale, design, methods and baseline demography of
participants of the Anglo-Scandinavian Cardiac Outcomes Trial.
ASCOT investigators. J Hypertension 2001; 19: 1139–1147.
18 Ziemer DC, Kolm P, Weintraub WS, Vaccarino V, Rhee MK,
Twombly JG et al. Glucose-independent, black-white differences in
hemoglobin A1c levels: a cross-sectional analysis of 2 studies. Ann
Intern Med 2010; 152: 770–777.
19 Avignon A, Radauceanu A, Monnier L. Nonfasting plasma glucose
is a better marker of diabetic control than fasting plasma glucose in
type 2 diabetes. Diabetes Care 1997; 20: 1822–1826.
20 Tillin T, Dhutia H, Chambers J, Malik I, Coady E, Mayet J et al. South
Asian men have different patterns of coronary artery disease when
compared with European men. Int J Cardiol 2008; 129: 406–413.
21 Park S, Barrett-Connor E, Wingard DL, Shan J, Edelstein S. GHb is
a better predictor of cardiovascular disease than fasting or
postchallenge plasma glucose in women without diabetes. The
Rancho Bernardo Study. Diabetes Care 1996; 19: 450–456.
22 Meigs JB, Larson MG, D’Agostino RB, Levy D, Clouse ME,
Nathan DM et al. Coronary artery calcification in type 2 diabetes
and insulin resistance: the framingham offspring study. Diabetes
Care 2002; 25: 1313–1319.
23 Selvin E, Steffes MW, Zhu H, Matsushita K, Wagenknecht L,
Pankow J et al. Glycated hemoglobin, diabetes, and cardiovascular
risk in nondiabetic adults. N Engl J Med 2010; 362: 800–811.
ª 2015 The Authors.
Diabetic Medicine published by John Wiley & Sons Ltd on behalf of Diabetes UK.
DIABETICMedicine
24 Gerstein HC, Anand S, Yi QL, Vuksan V, Lonn E, Teo K et al. The
relationship between dysglycemia and atherosclerosis in South
Asian, Chinese, and European individuals in Canada: a randomly
sampled cross-sectional study. Diabetes Care 2003; 26: 144–149.
25 Lee M, Saver JL, Hong KS, Song S, Chang KH, Ovbiagele B. Effect
of pre-diabetes on future risk of stroke: meta-analysis. BMJ 2012;
344: e3564.
26 Selvin E, Rawlings AM, Bergenstal RM, Coresh J, Brancati FL. No
racial differences in the association of glycated hemoglobin with
kidney disease and cardiovascular outcomes. Diabetes Care 2013;
36: 2995–3001.
27 National Institute for Health and Care Excellence. Preventing type
2 diabetes: risk identification and interventions for individuals at
high risk. NICE, 2012.
28 Ikehara S, Tabak AG, Akbaraly TN, Hulman A, Kivimaki M,
Forouhi NG, Iso H, Brunner EJ. Age trajectories of glycaemic traits
in non-diabetic South Asian and white individuals: the Whitehall II
cohort study. Diabetologia 2015; 58: 534–542.
Supporting Information
Additional Supporting Information may be found in the
online version of this article:
Table S1 Characteristics of participants in the Southall and
Brent Revisited study, by ethnicity and glycaemic status.
Table S2 Glycaemic status of participants in the Southall and
Brent Revisited study using American Diabetes Association
criteria for prediabetes [HbA1c ≥39 mmol/mol (5.7%) to
<48 mmol/mol (6.5%)], by ethnicity.
Table S3 Cardiovascular disease by ethnicity and glycaemic
status in the Southall and Brent Revisited study, using
American Diabetes Association criteria for prediabetes
[≥39 mmol/mol (5.7%) and <48 mmol/mol (6.5%)], by
ethnicity.
Table S4 Clinical or subclinical cardiovascular disease by
ethnicity and glycaemic status in the Southall and Brent
Revisited study.
Table S5 Multivariable models of cardiovascular disease by
ethnicity and glycaemic status; ADA prediabetes thresholds.
Table S6 Multivariable models of cardiovascular disease by
ethnicity and glycaemic status in the Southall and Brent
Revisited study, with further adjustment for medication use.
347